Development of Sensitive and Specific Determination Methods of Bioactive Compounds Based on Generation of High-Performance Antibodies

Abstract

Immunoassays enable the sensitive determination of various compounds and have been widely utilized in pharmaceutical and medical sciences. To develop practical assays, it is essential to obtain antibodies that capture the target analytes with high specificity and affinity. To date, we have generated high-performance antibodies and developed immunoassays for determining bioactive compounds, particularly focusing on haptens, such as steroids and synthetic drugs. In previous studies, we have produced specific anti-hapten antibodies by immunizing animals with reasonably prepared hapten–carrier conjugates. However, the resulting antibodies sometimes lacked sufficient affinity for a sensitive determination. Therefore, we challenged genetic engineering to produce artificially modified antibodies with improved affinity. Therein, native antibodies with insufficient affinities were converted into single-chain Fv fragments (scFvs), to which random point mutations were introduced to generate diverse scFv libraries. Mutated scFv species with increased affinities were selected and isolated with the aid of phage-display system combined with panning. Using this strategy, we obtained scFvs specific to several haptens, such as estradiol-17β (E₂) and cotinine, that show significantly improved affinity (K_a) than that of the parental scFv, enabling more sensitive enzyme-linked immunosorbent assays. However, the panning step often fails in straightforward selection and requires laborious trial-and-error work. Thus, we developed a “clonal array profiling (CAP)” system for more efficient isolation of the mutants with enhanced affinities, which successfully functioned generating multiple anti-cortisol scFvs with the K_a improved up to 63-fold and an anti-E₂ scFv with 372-fold larger K_a. In this study, we identified new strategies that allow for efficient site-directed mutagenesis to improve affinity. We expect that the engineered antibodies described here will open the door to next-generation immunoassays that will enable simpler and more reliable determination of bioactive compounds.

1. INTRODUCTION: AN IMMUNOASSAY MINI REVIEW

In 1959, Berson and Yalow invented “immunoassays,” a revolutionary analytical principle based on antigen–antibody reactions, which enabled us to quantify trace amounts of antigens (the compounds that are bound by antibodies) in biological fluids.^1–3) In their original reports, insulin was quantified via a competitive reaction with radioisotope-labeled insulin against an anti-insulin antibody (Fig. 1A-a), and thus the method was called “radioimmunoassay (RIA).” These achievements earned Yalow the Nobel prize in 1977.³⁾ Then, various modified methods have been developed to improve their feasibility and practicability.^4,5) Current immunoassays usually employ non-isotopic materials (e.g., enzymes and chemiluminescent dyes) to label antigens or antibodies to facilitate the monitoring of antigen–antibody reactions. Among them, “enzyme-linked immunosorbent assay (ELISA)”⁶⁾ might be the most standard format. Therein, either antigens or antibodies are labeled with enzymes, and the other is immobilized on solid phases, such as microplates or microbeads, to allow for quicker and automatized operations. Furthermore, a convenient solid-phase-based method, “immunochromatography,”⁷⁾ has recently been developed and is now used widely for on-site analysis and point-of-care testing.

Fig. 1. Schematic Illustration of the Principles of (A) Competitive and (B) Sandwich Immunoassay, and (C, D) Noncompetitive Determination of Haptens Using (C) Anti-Metatype Antibodies and (D) Anti-Idiotype Antibodies

(A) Competitive immunoassays are classified into the following two types: (a) the target antigen (haptens or macromolecules) and a fixed amount of antigen labeled with a signal-generating group, e.g., radioisotopes, enzymes, and chemiluminescent dyes, are competitively reacted against a limited number of antibody. The antibody-bound fraction of the labeled antigen (fraction B) is then separated from the free labeled antigen (fraction F) and the signal intensity of either fraction (usually B) is determined. (b) The target antigen (haptens or macromolecules) and a fixed amount of antigen (or its analog) immobilized on a solid phase, e.g., microplates and microbeads, are competitively reacted against a limited number of antibody (both monovalent fragments and bivalent antibodies are available) labeled with a signal-generating group. The liquid phase is then removed, and the signal intensity of the solid phase is determined. (B) The target antigen (macromolecules) is reacted with an excess amount of immobilized and labeled antibodies, simultaneously or sequentially, to form a sandwich-type immune complex. Then, the liquid phase is removed, and the signal intensity of the solid phase is determined. (C) The target hapten is captured by an immobilized anti-hapten antibody, and the resulting immune complex is then captured by an anti-metatype antibody labeled with a signal-generating group. Then, the liquid phase is removed, and the signal intensity of the solid phase is determined. An alternative system, in which anti-metatype antibody is immobilized and anti-hapten antibody is labeled, is available. (D) The target hapten is captured by an immobilized anti-hapten antibody, and the unreacted paratopes are “blocked” by reaction with the β-type anti-idiotype antibody (β-Id). Then, the variable regions of the anti-hapten antibody that captured the hapten and escaped from blockade with β-Id are selectively bound by the α-type anti-idiotype antibody (α-Id) labeled with a signal-generating group. Then, the liquid phase is removed and the signal intensity of the solid phase is determined. An alternative system, in which the α-Id is immobilized and anti-hapten antibody is labeled, is also available.

Immunoassays can be used to determine various materials. The requirement to fulfill as “analytes” is substantial solubility in aqueous media (where the antigen–antibody reactions proceed) and availability of relevant antibodies. Small molecules immunochemically classified as haptens (e.g., steroids and synthetic drugs)^5,8) are applicable as well as macromolecules that are immunogenic per se (e.g., proteins). Bacteria and viruses, which are large clusters composed of these macromolecules, are suitable analytes. Therefore, the use of immunoassays for clinical diagnostics in the early days was reasonable. However, the application has now been extended to various research fields, including the food industry⁹⁾ and environmental sciences.¹⁰⁾ Whereas haptens are determined almost solely by the competitive assay format (Fig. 1A), macromolecules are usually determined by the sandwich assay (Fig. 1B), a typical noncompetitive assay format using two types of antibodies, although the competitive assay format is also applicable.^5,8,11)

In immunoassays, antibodies [mostly immunoglobulin G (IgG) is used] (Fig. 2A) function as analytical reagents that recognize and capture analytes.^4,5) Before 1980, antisera obtained from animals immunized with analytes were typically used. However, antisera contain various antibody molecules, each derived from different B cell clones and shows different binding characteristics against the analyte (polyclonal antibodies). This limits the standardization of immunoassays because antisera with identical constituents are difficult to reproduce. B cell hybridoma technology, invented by Köhler and Milstein, eliminated this limitation by enabling the production of monoclonal antibodies (mAbs) derived from a single B cell.^5,12,13) Antibody-producing cells from immunized animals (mainly mice) are immortalized by fusion with myeloma cells and cloned to establish hybridoma cell lines, each secreting homogeneous antibody molecules with a unique primary structure and antigen-binding properties. Owing to these advantages, most modern immunoassays employ mAbs.

Fig. 2. Schematic Illustration of Antibodies and Their Fragments Discussed in This Review

(A) Typical native antibody used in immunoassays (IgG1 subclass is shown). The red bars that bind the H and L chains indicate disulfide bonds. The variable domains of the H and L chains (V_H and V_L) form paratopes that capture antigens. (B) A single-chain Fv fragment (scFv) that is prepared genetically by linking the V_H and V_L derived from a native antibody via a linker peptide. (C) A library of mutated scFvs (red stars indicate mutations) displayed on filamentous phage particles (scFv–phages) by fusion with a component of the phages (mainly the minor coat protein pIII). (D) The soluble form of the mutated scFvs with improved function. (E) Fab fragment obtained by the papain digestion of native antibodies.

Regardless of whether they are polyclonal or monoclonal in origin, the antigen-binding characteristics of the antibody essentially determine the analytical performance of the immunoassay such as specificity and sensitivity. However, it is not always easy to generate practical antibodies. For decades, we have developed immunoassay systems targeting bioactive compounds, particularly focusing on haptens of pharmaceutical interest, based on the de novo generation of practical antibodies. Here the author reviews our research results from the perspective of generating high-performance antibodies. This started from the classic approach of immunizing animals with reasonably designed immunogens and was then extended to genetic engineering of the native (animal-produced) antibodies to generate artificially modified antibody fragments with improved antigen-binding affinities (Figs. 2A–D).

2. GENERATION OF SPECIFIC ANTI-HAPTEN ANTIBODIES BASED ON THE DESIGN OF IMMUNOGENS AND THEIR APPLICATIONS

2.1. General Strategy for Generation of Anti-hapten Antibodies

A wide range of small molecules offer great value in clinical diagnosis, hygiene control, and environmental conservation. Compounds with a relative molecular mass (M_r) of less than 1000 usually function as haptens. Thus, the molecules themselves are not immunogenic, but once conjugated covalently to suitable macromolecules (carrier molecules), the resulting haptenic residues behave as antigenic determinants.^5,8,11) Animals immunized with hapten–carrier conjugates generate anti-hapten antibodies that bind not only to the conjugated residues but also to free (unmodified) hapten molecules. It is well accepted that antibodies elicited against hapten–carrier conjugates usually recognize partial structures of the hapten molecule that are remote from the position used for conjugation, whereas partial structures near the conjugation site are hardly recognized.^5,8) This tendency afforded us a strategy for designing hapten–carrier conjugates to generate antibodies showing a desirable recognition pattern: hapten molecules should be linked to carrier molecules via a site remote from the functional groups of the hapten differentiating it from possible cross-reacting analogs. Such conjugates should retain hapten residues to expose the characteristic functional groups outside the large carrier molecules and facilitate the approach of B cells equipped with surface immunoglobulins with desirable specificities, which are eventually released as antibodies available for immunoassay systems.

Based on this strategy, we produced novel antibodies specific to various haptens, e.g., steroid hormones,^14–16) metabolites of bile acids,^17–19) and vitamin D₃,^20–23) thyroxine,²⁴⁾ environmental pollutants (dioxins²⁵⁾ and nicotine metabolite²⁶⁾), controlled substances (cannabinoids,²⁷⁾ narcotics and psychotropics,^28,29) and stimulants³⁰⁾). Practical immunoassays and related immunochemical methods have been developed using the obtained antibodies. The topics selected are as follows.

2.2. Generation of Antibodies Specific to Vitamin D₃ Metabolites

Vitamin D₃ is taken through foods or supplied endogenously via 7-dehydrocholesterol, and is converted into 1α,25-dihydroxyvitamin D₃ [1,25(OH)₂D₃] (Fig. 3A-a), which participates in the regulation of calcium and phosphorus metabolism. The determination of blood 1,25(OH)₂D₃ levels is useful for diagnosing renal and parathyroid disorders. Previously, competitive binding assays using vitamin D receptor (VDR) derived from the calf thymus or chick intestine were used for this purpose.^31,32) Although the VDR exhibited extremely high affinity to 1,25(OH)₂D₃ [the affinity constant (K_a) exceeded 10¹⁰ M⁻¹], careful storage was necessary to maintain binding activity.

Fig. 3. Summary of the Production and Characterization of Antibodies Specific to 1,25(OH)₂D₃

(A) Structures of (a) vitamin D₃ and 1,25(OH)₂D₃ (A, C, and D rings and positions 1, 3, 11, and 25 in the vitamin D skeleton are shown), and (b) immunogenic conjugates in which 1,25(OH)₂D₃ molecules are linked to a carrier macromolecule via position 3 (left) or 11 (right). (B) Synthesis of 11α-hydroxy-1,25(OH)₂D₃ hemiglutarate for linking to carrier macromolecules via the position 11. The reaction conditions and yields are described in ref. 34. (C) Cross-reactivities and affinity constants (K_a) of anti-1,25(OH)₂D₃ antisera Ab-1 and -3 (obtained from different rabbit individuals)²¹⁾ and mAbs M114 and M129 (both mouse IgG2a κ; generated from different hybridoma clones).²²⁾ Cross-reactivities were evaluated based on the 50% displacement method.⁹⁸⁾ The structures of 1,25(OH)₂D₃ analogs tested here are shown, in which the partial structures differed from 1,25(OH)₂D₃ are highlighted in pink. Abbreviation: 25,26(OH)₂D₃, (25S)-25,26-dihydroxyvitamin D₃. K_a values were determined using Scatchard analysis.⁹⁹⁾

Because antibodies are more stable and easier to handle than VDR, immunoassays were expected to be an alternative method. However, several RIAs developed at that time lacked specificity; thus, complicated pretreatments of blood specimens were necessary.³³⁾ This was because the antisera used were produced against immunogens in which 1,25(OH)₂D₃ molecules were linked to carrier molecules via position 3 (Fig. 3A-b; left). According to the tendencies described above, these immunogens can induce antibodies that are cross-reactive with the 1,25(OH)₂D₃ analogs modified on the A ring. To generate specific antibodies, 1,25(OH)₂D₃ molecules should be linked via a site that is remote from both the A ring and the side chain on the D ring. Position 11 was a suitable site for this purpose (Fig. 3A-b; right) but was not used because of the difficulty in chemically modifying the C ring of 1,25(OH)₂D₃.

Thus, we successfully synthesized 11α-hemiglutaryloxy-1,25(OH)₂D₃ in 31 steps, starting from a microbiologically 11-hydroxylated steroid (Fig. 3B), which was then linked to bovine serum albumin (BSA; a commonly used carrier molecule) via amide bond formation³⁴⁾ (Fig. 3A-b). By immunizing rabbits and mice with the resulting conjugate, antisera²¹⁾ and mAbs²²⁾ against 1,25(OH)₂D₃ were generated. As expected, both antibodies discriminated the analogs modified on the A ring (e.g., 25-hydroxyvitamin D₃ [25(OH)D₃]) as well as those modified on the side chain (e.g., 1α,25-dihydroxyvitamin D₂ [1,25(OH)₂D₂]), as shown with the cross-reactivities (Fig. 3C) in a competitive RIA system using a [³H]-labeled 1,25(OH)₂D₃ (Fig. 1A-a). The cross-reactivities with 1,25(OH)₂D₂ and (24R)-1α,24,25-trihydroxyvitamin D₃ [1,24,25(OH)₃D₃] (for the antisera) were significantly lower than that found with the VDR, which demonstrated the advantage of antibodies over other native binding proteins, such that “custom-made specificity” is achievable by the design of immunogens. Although the affinity to 1,25(OH)₂D₃ did not exceed that of the VDR (Fig. 3C), the genetic engineering described later might generate mutated antibodies with satisfactorily enhanced affinities. The antisera were used to develop an immunoaffinity extraction system for the pretreatment of serum specimens.²¹⁾ We note that specific antibodies against the related metabolites, 25(OH)D₃²⁰⁾ and (24R)-24,25-dihydroxyvitamin D₃ [24,25(OH)₂D₃]²³⁾ (Fig. 3C), were similarly produced using immunogens in which these metabolites were linked via position 11.

2.3. Generation of Antibodies Group-Specific to Polar Conjugated Bile Acid Metabolites

Bile acids, a group of steroids with a carboxylated side chain, are biosynthesized in the liver, the majority of which are amidated with glycine or taurine before being excreted into bile. It was shown that, in hepatobiliary diseases, further conjugation is promoted to produce highly polar metabolites. For example, 7-N-acetylglucosaminides of nonamidated, glycine- and taurine-amidated ursodeoxycholic acid (UDCA 7-NAGs) (Fig. 4A) were found in the urine of a patient with primary biliary cirrhosis after the administration of ursodeoxycholic acid (UDCA).^35,36) UDCA is a minor component of endogenous bile acid and is also used as a medicine for hepatobiliary diseases. Therefore, the total amount of UDCA 7-NAGs was expected to be a useful diagnostic index.

Fig. 4. Summary of the Production and Characterization of Antibodies Group-Specific to Polar Conjugated Bile Acid Metabolites

(A) Structures of UDCA and its related compounds. (B) Preparation of the immunogenic conjugate for generating antibodies group-specific to UDCA 7-NAGs. Nonamidated UDCA was linked to BSA via the N-hydroxysuccinimide ester method.¹⁰⁰⁾ (C) Dose–response curves of UDCA 7-NAGs in the competitive ELISA (Fig. 1A-a) using the microplates coated with the mAb group-specific to UDCA 7-NAGs (Ab-#8; mouse IgG1 κ).¹⁷⁾ (D) Serial dilution study for measuring a urine specimen with and without pretreatment.

Immunoassays are usually used for monitoring a single analyte but are also able to determine the sum of structurally related compounds. To determine the UDCA 7-NAGs altogether, we first generated group-specific mAbs that equally bind to UDCA 7-NAGs, irrespective of the amidation status.¹⁷⁾ A suitable immunogen was easily obtained by linking nonamidated UDCA 7-NAG directly to BSA via its own carboxy group (Fig. 4B). In the resulting conjugate, the side-chain-derived structure should be masked by the carrier, whereas the 7-NAG group should be well exposed. Mice were immunized with the conjugate, their splenocytes were fused with P3/NS1/1-Ag4-1 myeloma cells, and a hybridoma clone secreting the desirable antibody was established. In a competitive ELISA system using β-galactosidase-labeled nonamidated UDCA 7-NAG (Fig. 1A-a), this mAb showed the desirable group-specificity, as shown by the similar and sensitive dose–response curves for each amidated status of UDCA 7-NAG, which covered approx. 1–200 pg/assay as a measurable range (Fig. 4C). In contrast, UDCA analogs lacking the 7-NAG group [nonamidated UDCA, glycine-amidated UDCA, and nonamidated UDCA 3-sulfate (3-Sul); Fig. 4A] showed negligible cross-reactivity (<0.01%). Consequently, this ELISA enabled the direct measurement (requiring no pretreatment) of urinary UDCA 7-NAGs. The assay values were validated by serial dilution tests and a comparison of the values before and after extraction using C18-cartridges (Fig. 4D).

Sulfation of the 3-hydroxy group was considered to be a major metabolic pathway of UDCA in both healthy subjects and patients with cholestatic and chronic liver diseases.³⁷⁾ According to the same strategy, we also generated an mAb group-specific to 3-sulfated UDCAs (UDCA-Suls) (Fig. 4A) for use in ELISA.¹⁸⁾ A practical mAb targeting sulfated metabolites of lithocholic acid (LCA-Suls) (Fig. 4A), which were expected as a useful index of liver function, was also established and utilized for developing an ELISA for diagnosing damaged liver functions.¹⁹⁾

2.4. Generation of Antibodies Specific to Hallucinogenic Alkaloids

Psilocybin (Pyb) and psilocin (Psi) are tryptamine alkaloids present in hallucinogenic mushrooms^38,39) (Fig. 5A). Following administration, Pyb is rapidly dephosphorylated to generate Psi, which is pharmacologically active agent. The possession and use of Pyb and Psi are prohibited in many countries. Immunoassays of Pyb and Psi were expected to be useful for examining these compounds and suspicious mushrooms. However, practical antibodies against Pyb or Psi were unavailable.

Fig. 5. Summary of Production and Characterization of Antibodies Specific to Hallucinogenic Alkaloids (reproduced from ref. 29 with partial modifications with permission from the Elsevier)

(A) Structures of Pyb, Psi, and their analogs with their cross-reactivities (the 50% displacement methods)⁹⁸⁾ in the competitive ELISAs (Fig. 1A-b) using anti-Pyb mAb (Pyb#74; mouse IgG2b κ) and anti-Psi mAb (Psi#105; mouse IgG3 κ).²⁹⁾ (B) Preparation of the immunogenic conjugates Pyb–BSA and Psi–KLH (KLH, keyhole limpet hemocyanin). The reaction conditions for each step are described in ref. 29. (C) ELISA dose–response curves of Pyb and Psi with midpoint values (the mass of analyte required for 50% displacement in the dose–response curves). Vertical bars indicate the S.D. of intra-assay variances (n = 4).

Thus, we generated two mAbs specific to Pyb or Psi and developed ELISAs.²⁹⁾ Pyb and Psi molecules were linked to carrier molecules by modifying their side chains (Fig. 5B). Spleen cells from mice immunized with the conjugates were fused with the myeloma cells and hybridoma clones secreting anti-Pyb and anti-Psi mAbs were established. These mAbs were applied to competitive ELISAs that used microplates with immobilized Pyb or Psi residues (Fig. 1A-b). The measurable ranges were approx. 0.20–20 and 0.040–2.0 µg/assay and the limit of detection (LOD), defined as the analyte amount that provided bound signals two standard deviation (S.D.) below the average (n = 10) of the signals at zero concentration, was 0.14 and 0.029 µg/assay for Pyb and Psi, respectively (Fig. 5C). Related tryptamines were satisfactorily discriminated, as exemplified by their cross-reactivity (Fig. 5A). The Pyb and Psi contents in a dried powder of the hallucinogenic mushroom, Psilocybe cubensis, were determined to be 0.39 and 0.32% (w/w), respectively, which were within the ranges reported so far.

2.5. Generation of Antibodies Specific to Derivatized Stimulants

Methamphetamine (MAP) and amphetamine (AP) are powerful central nervous system stimulants, and their abuse is a serious problem globally.⁴⁰⁾ Because (S)-enantiomers of these compounds (Fig. 6A-a) are much more potent than (R)-antipodes,⁴¹⁾ (S)-enantiomers or racemic compounds have been abused.⁴²⁾ Group-specific immunoassays for detecting (S)-MAP and (S)-AP are useful for regulating their illicit circulation. However, it has been difficult to generate antibodies that react equally to these stimulants because of the significant differences between their characteristic structures: i.e., N-methylamino (MAP) and amino (AP) groups. In addition, their low molecular mass (M_r<150) hampers the generation of high-affinity antibodies.

Fig. 6. Summary of the Production and Characterization of Antibodies Specific to the Derivatized Stimulants (reproduced from ref. 30 with partial modifications with permission from the Royal Society of Chemistry)

(A) Structures of (a) (S)-MAP (R = CH₃) and (S)-AP (R = H), (b) their Teoc derivatives, and (c) immunogenic conjugates used for antibody production. (B) Dose–response curves of Teoc-(S)-MAP and Teoc-(S)-AP standards with the midpoint and LOD values [vertical bars indicate the S.D. (n = 4)] and cross-reactivities (the 50% displacement method)⁹⁸⁾ in the ELISA using the mAb (mAb-tAP#5-7; mouse IgG2b λ) derived from the mouse immunized with Teoc-(S)-AP–BSA. (C) Detection of (S)-MAP hydrochloride and (S)-AP sulfate by immunochromatography. These stimulants (0.010, 0.10, or 1.0 mg as free bases) were derivatized and applied to immunochromatography devices equipped with a conjugate pad containing mAb-coated gold nanoparticles and a test line coated with the Teoc-(S)-MAP–BSA conjugate. The values on the test lines represent the absorbance (mAU values) determined using an immunochromatogram reader.

To overcome these problems, we converted (S)-MAP and (S)-AP into their 2-(trimethylsilyl)ethyl carbamates,⁴³⁾ i.e., Teoc-(S)-MAP and Teoc-(S)-AP (Fig. 6A-b), respectively, to increase their molecular mass and simultaneously mask their structural differences. To generate an mAb that reacts equally to these Teoc-derivatized forms, mice were immunized with BSA conjugates of Teoc-(S)-MAP or Teoc-(S)-AP (Fig. 6A-c), and their spleen cells were fused as usual. As expected, we obtained an mAb that provided almost overlapping dose–response curves for Teoc-(S)-MAP and Teoc-(S)-AP (Fig. 6B) in a competitive ELISA using microplates immobilizing Teoc-(S)-MAP residues³⁰⁾ (Fig. 1A-b). The cross-reactivity of related Teoc derivatives demonstrated that the mAb recognized the (S)-MAP/(S)-AP structures as well as the Teoc group (Fig. 6B).

To evaluate the discrimination ability of the derivatization/immunoassay system, sample powders [salts of (S)-MAP/(S)-AP or related amines, fixed at 1.0 mg as a free base] were incubated with Teoc-O-succinimidyl⁴³⁾ for 5 min at room temperature. Then, a fixed portion of the reaction mixture was subjected to the ELISA using Teoc-(S)-MAP as the calibration standard to assess “Teoc-(S)-MAP equivalent values.” Both (S)-MAP and (S)-AP were assessed as >100 µg/assay (indicating that approx. 30% of the compounds were derivatized), distinguishing from (1R,2S)-ephedrine, (S)-methylenedioxymethamphetamine, tyramine, dopamine, β-alanine (assessed as below the LOD), and the antipodes (R)-MAP and (R)-AP (assessed as 19 and 25 µg/assay, respectively). This “derivatization-assisted ELISA” detected as little as 10 µg of (S)-MAP or (S)-AP. Immunochromatography devices were also developed using gold nanoparticles coated with the mAb, with which 0.10 mg of (S)-MAP and (S)-AP were detected with the naked eye (Fig. 6C). We expected that the present immunoassays may be useful for the detection of (S)-MAP/(S)-AP in the early-stage screening of suspicious substances.

3. “ANTIBODY-BREEDING” TO IMPROVE IMMUNOASSAY SENSITIVITIES

3.1. Strategies for Improving Immunoassay Sensitivities

As mentioned above, haptens were almost exclusively determined using competitive immunoassay formats (Fig. 1A). Theoretical dose–response curves drawn under ideal conditions (where analytes and labeled analytes equally react to antibodies) show that the antibodies with a higher affinity (with a larger K_a value) enable the assays with a higher sensitivity⁴⁴⁾ (Fig. 7A). This relationship was also recognized in “real ELISAs” that we developed (Fig. 7B). The five mAbs used in these ELISAs were prepared using similar immunization/cell fusion schedules: immunization was repeated three or four times to promote in vivo affinity maturation. However, the K_as of resulting antibodies differed significantly. While the ELISA of UDCA 7-NAGs (see Section 2.3) using the mAb with the largest K_a (5.4 × 10⁹ M⁻¹) allowed for satisfactorily sensitive measurement as shown with the picogram-order midpoint (0.017 ng/assay), the ELISA of Pyb (see Section 2.4) using the mAb with the smallest K_a (9.1 × 10⁵ M⁻¹) was approx. 10⁵-fold less sensitive as calculated based on the midpoint (1370 ng/assay). Another important aspect shown in Fig. 7B is that the K_a values of the anti-hapten antibodies prepared by immunization-based procedures rarely exceeded 10¹⁰ (M⁻¹) range. These affinity ceilings lead to sensitivity ceilings in hapten determination. It is often difficult to generate high-affinity antibodies when the target hapten has a particularly small molecular mass^45,46) (as a general criterion, M_r <300).

Fig. 7. Effect of Antigen-Binding Affinity of Antibodies on Immunoassay Sensitivities

(A) Theoretical dose–response curves of competitive immunoassays based on the principle shown in Fig. 1A-a. In these simulations, 25(OH)D₃ (M_r 400.6) (Fig. 3C) was chosen as the model analyte.⁴⁴⁾ Dose–response curves in an assumed RIA system, in which the competitive reactions were performed in 500 µL of medium using 2 pM ¹²⁵I-labeled 25(OH)D₃ as the labeled hapten and anti-25(OH)D₃ antibodies with five different K_a values [each used to allow for 50% binding of added ¹²⁵I-labeled 25(OH)D₃ at zero concentration (B₀)], were constructed. The B/B₀ percentage was calculated using the following equation, where X′, R′, and A′ (each in pM) are the concentrations of the unlabeled hapten, antibody binding site, and labeled hapten, respectively. K_d (pM) is the dissociation constant (=1/K_a) of the antigen–antibody reaction.

(B) Dose–response curves of ELISAs for UDCA 7-NAGs (see Section 2.3), hydrochlorides of (R)- and (S)-ketamine,²⁸⁾ Pyb and Psi (see Section 2.4) using hybridoma-derived mAbs. The values in orange are the midpoints (ng/assay). The name of these mAbs in the original articles, their K_a values, and the assay type employed, are listed. The K_a values of mAb-Psi#105 and Pyb#74 are “apparent K_a” determined for bivalent antibodies against hapten-immobilized solid phases.

Sandwich immunoassays usually allow for more sensitive determinations than competitive immunoassays.^11,44) Therein, target compounds are captured in the solid phase via excess antibodies and then detected via the reaction with excess antibodies labeled with a signal-generating group (Fig. 1B). The use of excess antibodies promotes the antigen–antibody reaction so that even trace amounts of the analyte are captured and detected. However, this principle requires that the analytes have at least two epitopes in a single molecule, and thus cannot be applied directly to most haptens. To overcome this limitation, several noncompetitive assay formats applicable to haptens have been devised.¹¹⁾ The most promising method was semi-sandwich-type assays employing “anti-metatype antibodies,” which are specific to a particular complex of hapten and anti-hapten antibody (Fig. 1C). In the 1990s, some applications were reported in which Δ⁹-tetrahydrocannabinol (THC),⁴⁷⁾ digoxin,⁴⁸⁾ and angiotensin II⁴⁹⁾ were determined. The reported LOD of the digoxin and angiotensin II assays were 30 ng/L and 1 pg/mL, respectively. These assays used anti-metatype mAbs prepared by conventional immunization-based methods. Subsequently, genetically engineered antibody fragments showing anti-metatype specificities were successfully employed to determine haptens, e.g., morphine,⁵⁰⁾ estradiol,⁵¹⁾ and aldosterone.⁵²⁾ However, the production of practical anti-metatype antibodies is still difficult, and there seems to be little prospect of discovering reliable strategies. One reason for this lies in the general feature of hapten–anti-hapten interactions; that is, hapten molecules tend to be almost entirely buried in the paratopes of antibodies. The combinatorial use of α- and β-type anti-idiotype antibodies with anti-hapten antibodies also enables the noncompetitive determination of haptens^53–56) (Fig. 1D). The α- and β-type anti-idiotype antibodies are “the second antibodies” that bind to the variable region of a particular antibody in noncompetitive and competitive manners with relevant antigens, respectively. Based on this principle, we succeeded in determining 11-deoxycortisol with approx. 1000-fold improved sensitivity (20 amol of the analyte was detectable).⁵⁷⁾ However, anti-idiotype-based assays require three kinds of antibodies and rather complicated procedures, although the generation of anti-idiotype antibodies is easier than that of anti-metatype antibodies.

Taking these situations into account, we were attracted to the potential of “antibody engineering” (genetic engineering of antibody molecules), which advanced extensively since the 1990s.^58–64) This technology enabled in vitro affinity maturation, i.e., in vitro generation of artificially modified antibodies that show higher antigen-binding affinities compared with the native antibodies. We called this “antibody-breeding,” and expected to make a breakthrough for increasing immunoassay sensitivities. However, most studies that have been performed on this subject have aimed at developing therapeutic antibodies. Applications in analytical chemistry have rarely been reported. Therefore, we challenged the breeding of native antibodies to open a new avenue for more sensitive hapten immunoassays.

3.2. Antibody-Breeding for More Sensitive Hapten Immunoassays

From among a variety of approaches, we chose a strategy that was established based on the achievements of Winter and Smith.^58,60,64) Therein, a native mAb to be improved (parental antibody) is converted into single-chain Fv fragment (scFv)^65,66) (Figs. 2A, B). Subsequently or simultaneously, a diverse population of genetically randomized scFv species (scFv library) is produced, from which mutated scFv species showing increased affinities are selected and isolated with the aid of phage display technique^64,67,68) (Figs. 2C, D). The experimental protocol contained the following steps (Fig. 2): (1) the V_H and V_L genes each encoding the variable domain of the heavy (H) and light (L) chains (the V_H and V_L) of a parental antibody were prepared; (2) the gene encoding scFv was assembled by linking the V_H and V_L genes via a linker sequence, followed by introduction of random point mutations via the error-prone PCR⁶⁹⁾; (3) Escherichia coli (E. coli; XL1-Blue strain) was transformed with the randomized scFv genes and then infected with VCSM13 helper phage to express scFv proteins as a fused form with M13 phage particles (the phage display) to produce a library of “scFv–phages” containing >10⁵ species; (4) scFv–phage species that gained improved affinities were selected and isolated from the library via the panning (as it were, affinity extraction) against target haptens immobilized on polystyrene tubes; and (5) the mutated scFv with the improved affinity (m-scFv) recovered was converted into “soluble protein,” which is not linked with phage particles and characterized in ELISA systems.

We applied this strategy to improve the affinities of four hybridoma-derived mAbs, each specific to estradiol-17β (E₂),^70,71) cotinine,⁷²⁾ cortisol,⁷³⁾ or THC²⁷⁾ (Fig. 8). It should be appended that (1′) scFv molecules were assembled in the direction of V_H-linker-V_L (Fig. 2B, left) employing the linker sequence of VSS(GGGGS)₃T (for the cortisol-specific scFv) or (GGGGS)₃ (for the other scFvs) [G, glycine; S, serine; T, threonine; and V, valine]; (2′) regarding the E₂-specific scFv, mutagenesis (preparation of a mutated scFv library followed by selection of improved species) was repeated three times in order to “overwrite” mutations; and (3′) regarding the cortisol-specific scFv, either the V_H or V_L was mutagenized to explore which domain is more important as the target for increasing the affinity.

Fig. 8. Summary of Antibody-Breeding Performed for scFvs against (A) E₂, (B) Cotinine, (C) Cortisol, and (D) THC [reproduced from Sci. Rep., 10, 4807 (2020), Springer Nature, with partial modifications]

Typical dose–response curves for competitive ELISAs (Fig. 1A-b) using wt-scFv (blue), native Fab [blue, shown only in (A)], and m-scFvs (pink) are shown. These m-scFvs were named in the original articles as (A) scFv#M3rd,⁷⁴⁾ (B) scFv#m1-54,⁷²⁾ (C) scFv#m1-L10,⁷³⁾ and (D) scFv#m1-36,²⁷⁾ respectively. Vertical bars indicate the S.D. (n = 4). The magnitude of the improvements in the assay sensitivity (calculated based on the decrease in the midpoint values) is also shown. The primary structures of the m-scFvs were deduced from their nucleotide sequences. The numbering of the V_H and V_L and the positions of the CDRs are based on the Kabat definition.⁷⁵⁾ V_H-CDR1, 2, 3, and V_L-CDR1, 2, and 3 are abbreviated as H1, H2, H3, L1, L2, and L3, respectively. “FLAG” means the FLAG tag (DYKDDDDK) [K, lysine]. The introduced amino acid substitutions are denoted by stars and one-letter codes. Regarding m-scFv against E₂, the four substitutions shown in orange one-letter codes are particularly important for affinity enhancement (see text). The K_a values (M⁻¹) of antibody fragments are as follows: (A) wt-scFv, 8.6 × 10⁷; wt-Fab, 5.2 × 10⁷; m-scFv, 1.2 × 10¹⁰; (B) wt-scFv, 2.7 × 10⁷; m-scFv, 1.2 × 10⁹; (C) wt-scFv, 3.4 × 10⁸; m-scFv, 1.2 × 10¹⁰; and (D) wt-scFv, 1.1 × 10⁷; m-scFv, 1.1 × 10⁸, respectively. The K_a values for (A) and (C) were determined by Scatchard analysis.⁹⁹⁾ For scFvs for (B) and (D), surface plasmon resonance and biolayer interferometry sensors were used, respectively.

As shown in Fig. 8, we isolated m-scFvs that showed significantly improved affinity (K_a) in all trials. The number of amino acid substitutions introduced varied widely: i.e., 11, 5, 3, and 1 position(s) for scFvs specific to E₂, cotinine, cortisol, and THC, respectively. Regarding the E₂-specific m-scFv, we revealed that the 4 substitutions among the 11 are particularly important (discussed later)⁷⁴⁾ (Fig. 8A). Furthermore, the most critical substitution was estimated to be L100gQ, i.e., the substitution from leucine (L) to glutamine (Q) that occurred at the position 100 g in the V_H, which was included in the complementarity-determining region (CDR) 3.⁷⁵⁾ These findings would be reasonable considering the common understanding that the V_H-CDR3 sequences often play a crucial role in antigen-binding interactions by virtue of their most diversified structures compared to other CDRs.^76,77)

The K_a values increased by 140-, 44-, 35-, and 10-fold compared to the wild-type scFv (wt-scFv; the scFv combining the V_H and V_L sequences in a parental mAb without modifications) for E₂, cotinine, cortisol, and THC, respectively. Remarkably, the THC-specific m-scFv gained a 10-fold higher affinity via only a single substitution (S50T) in the V_H-CDR2: in other words, the addition of only a single methyl group in the scFv paratope²⁷⁾ (Fig. 8D). These enhanced affinities enabled the m-scFvs to develop competitive ELISAs (Fig. 1A-b) with higher sensitivities than the relevant wt-scFvs by 5-, 100-, 7-, and 3-fold (based on the comparison of the midpoints) (Fig. 8). The E₂ ELISA with the m-scFv was 33-fold more sensitive than the ELISA using the Fab fragment prepared from the parental mAb (wt-Fab).

These results suggested that “antibody-breeding” is a drastic and universal approach for improving the sensitivity of hapten immunoassays. In the cotinine ELISA, the use of m-scFv showing a 44-fold increased affinity (K_a >10⁹ M⁻¹) resulted in a 100-fold improved sensitivity, enabling the measurement of urinary cotinine levels for detecting passive smoke exposure (Fig. 8B). The LOD of this ELISA (8.4 pg/assay) corresponded to 0.17 ng/mL urinary cotinine. This is remarkable because the hybridoma methods hardly generated anti-cotinine mAbs with practical affinities due to the low M_r of cotinine (176.2).⁷²⁾ In fact, we previously produced hybridoma-based anti-cotinine mAbs but their K_a did not exceed 10⁷ M⁻¹.²⁶⁾ Regarding anti-cortisol scFv, the m-scFv we obtained was the species combining the parental V_H and a mutated V_L domain⁷³⁾ (Fig. 8C). This mutated scFv was improved for not only affinity but also specificity as shown by the decreased cross-reactivities with cortisone (45→17%), prednisolone (18→3.3%), and prednisone (0.78→0.31%) compared with the wt-scFv. These results suggest that the antibody-breeding approach is also promising for improving assay specificity. It should be noted that ELISA systems using the m-scFvs targeting E₂, cotinine, and cortisol were validated for their applicability to clinical specimens.

3.3. Problems to Be Solved for More Efficient Antibody-Breeding

The antibody-breeding has proven useful in improving the sensitivity of competitive immunoassays targeting haptens. It should be noted that high-affinity antibodies are similarly advantageous for developing the sandwich assays of macromolecules. However, much labor was needed until “the best mutant” (m-scFvs shown in Fig. 8) was discovered: we had to examine numerous scFv species (in some cases, >1000 species), which were selected via the panning.⁷⁸⁾ In fact, panning, though it has been used as the standard method in phage-display-based antibody engineering, often failed to straightforwardly identify improved scFv species. We estimated that one of the major causes of this problem is competition with a large excess of undesirable mutants with weaker affinities against a limited number of haptens immobilized on solid phases.⁷⁸⁾ Bias for propagation of transformants (scFv-gene-introduced E. coli cells), as well as for replication of scFv–phages, might also prevent successful selection.

Such limitations should be overcome by screening individual scFv–phage clones generated from the initial bacterial libraries that maintain the original diversity. Thus, we devised “clonal array profiling of scFv-displaying phages (CAP),” an efficient and robust system for discovering rare affinity-matured scFvs, which is described in the next chapter.

4. “CAP” FOR MORE EFFICIENT ANTIBODY-BREEDING

4.1. Construction of the CAP System

Our CAP system contained the following steps⁷⁸⁾ (Fig. 9): (1) E. coli (TG1 strain) was transformed with the scFv gene library to be explored; (2) and then immediately spread on agar plates so as not to disturb their original diversity; (3) the colonies grown on agar were transferred individually into microwells in 96-well plates pre-coated with the target antigen and filled with a liquid medium containing KM13 helper phage; (4) the plates were incubated to allow scFv–phages to propagate and, if they are antigen-specific with enough affinity, to be captured on the microwells; (5) the captured phages were detected with a bioluminescent assay using an in-house-prepared fusion protein linking an scFv against M13 phage and Gaussia (or NanoLuc) luciferase^24,78,79); (6) scFv–phages that generated strong luminescence were recovered from the microwells by acidic treatment (pH 2.2) [method a in Fig. 9(6)]; and (7) the recovered scFv–phages were converted into the soluble scFvs and characterized in ELISA systems.

Fig. 9. Schematic Illustration of the CAP System

The procedures for the steps (1)–(7) are described in the text.

This CAP approach might be regarded as a return to a “brute force screening” from a “theoretically rational selection” (panning). However, we were convinced that CAP would be a more straightforward and robust method to enable the high-throughput isolation of improved species.

4.2. Advantages of CAP over Conventional Panning

To assess the advantages of CAP, we performed a comparative study in which an anti-cortisol scFv library, newly prepared via error-prone-PCR covering the entire scFv sequence (initial transformants generated approx. 10⁵ colonies), was subjected to CAP and conventional panning in a parallel manner. The parental scFv used here (Fig. 10A) was the same as the anti-cortisol wt-scFv used in Section 3.2.⁷⁸⁾ The CAP was operated twice, each submitting approx. 3% of the library (i.e., 9400 transformant colonies), owing to the limited capacity for manual handling of colonies. In contrast, 100% of the library was subjected to three rounds of panning using polystyrene tubes coated with the cortisol–BSA conjugate. These panning experiments were carried out using three protocols that differed in the conditions for binding (cortisol concentrations in the solid phase), washing, and elution of scFv–phages.

Fig. 10. Comparison of the CAP-Derived and Panning-Derived scFv Mutants with Improved Affinity [reproduced from Sci. Rep., 10, 14103 (2020), Springer Nature, with partial modifications]

In the left half, schematic illustrations of the primary structures and the K_a values of the (A) wt-scFv, (B) CAP-derived, and (C) panning-derived scFv mutants are shown. The primary structures were deduced and are illustrated as shown in Fig. 8. Stars and arrowheads denote substitutions and insertions, respectively. Regarding the scFv#m1-2, an amber codon was inserted between the codons for positions 6 (Q) and 7 (P) that was readthrough as Q (confirmed by Edman sequencing). Details of the substitutions in other scFvs are described in ref. 78. The K_a values were determined by the Scatchard analysis⁹⁹⁾ (but with a slightly different condition from that used in Fig. 8C). In the right half, typical dose–response curves for cortisol in competitive ELISAs (Fig. 1A-b) generated using the (B′) CAP-derived and (C′) panning-derived mutants are shown with the midpoint values, comparing with those using the wt-scFv. Vertical bars indicate the S.D. (n = 4).

Despite the throughput handicap, CAP was evidently superior. As shown in Fig. 10B, CAP provided eight mutated scFvs with K_a values larger than 5 × 10⁹ M⁻¹, which were 14–63-fold improved over the wt-scFv (K_a = 3.8 × 10⁸ M⁻¹). From the panning, we obtained seven mutants with improved K_a, however, the largest value of which (1.9 × 10⁹ M⁻¹) was only 5.0-fold larger than that of the wt-scFv (Fig. 10C).

It should be noted that, during about 30 years after antibody engineering was established, less than 10 studies have succeeded in practical affinity maturation of anti-hapten antibodies: in which mutants with a K_a of over 10⁹ M⁻¹ were generated after over 4-fold improvement, and the mutants were reactive against free (not immobilized) hapten molecules.^44,78) CAP should readily be applicable for laboratories with standard equipment and should enable overcoming the longstanding trouble in isolating rare and improved antibody species. In the future, CAP might gain extensive working power by automatization using robotic colony pickers, which would enable the screening of over 10⁶-size libraries. If we processed the entire library discussed above, 267 [=8 × (100/3)] types of similarly-improved mutants might have been obtained.

4.3. Analytical Performance of the CAP-Derived Antibody Fragments

Cortisol is used as an index for the function of the hypothalamic–pituitary–adrenal axis⁸⁰⁾ and thus, practical anti-cortisol antibodies are in great demand. However, only a few publications have demonstrated the production of mAbs capable of targeting serum or urinary cortisol.^14,81,82) The CAP-derived five anti-cortisol scFvs with the K_a exceeded 10¹⁰ (M⁻¹) enabled the competitive ELISA (Fig. 1A-b) with dramatically enhanced sensitivities, as shown by 11–25-fold smaller midpoints (31–71 pg/assay) than that of the wt-scFv (768 pg/assay) (Fig. 10B′). Contrary, the panning-derived mutants resulted in only poor improvements, in which the decrease in the midpoints did not exceed 2.3-fold (Fig. 10C′). The LODs of the ELISAs using the four CAP-derived mutants that exhibited midpoints of less than 50 pg/assay (scFv#m1-10, -18, #m2-91, and -97) was 5.5–11 pg/assay, which corresponds to 1.1–2.2 ng/mL serum cortisol levels (assuming that serum specimens could be directly applied by diluting 10-fold). Considering the normal levels of serum cortisol (10–250 ng/mL),⁸³⁾ we estimated that these ELISA systems are sufficiently sensitive for diagnostic applications. The specificity of the assays was also acceptable in cross-reaction studies.⁷⁸⁾

4.4. Structural Aspects for Affinity Enhancements

We were interested in the structural aspects that could explain such differences in the potential to improve the affinity between CAP-derived and panning-derived mutants. Regarding panning-derived scFvs, the number of amino acid substitutions introduced in the entire molecule ranged from one to seven with an average of 4.1 (Fig. 10C). The proportions of these substitutions in the V_H and CDRs, which are recognized to play important roles in the interaction with antigens, were 40 and 24%, respectively (substitutions in the linker were excluded). However, the CAP-derived mutants were quite different (Fig. 10B); substitutions and insertions were fewer (the average number was 2.3), but the proportions in the V_H (81%) and CDRs (31%) were higher than those found in the panning-derived mutants.

Further notable aspect regarding the CAP-derived mutants was that only a single substitution (scFv#m1-10, -18, #m2-91, and -183) or a single insertion (scFv#m1-2) achieved the 16–37-fold larger K_as (Fig. 10B). Moreover, except for scFv#m1-18, these scFvs were all modified at their V_H framework region (FR) 1 (V_H-FR1). The V_H-FR1 is defined as the partial structure covering the amino acids at positions 1–30, located at the N-terminus of the V_H domain⁷⁵⁾ (Fig. 10A). Three mutants with multiple substitutions (scFv#m1-7, #m2-4, and -97) also substituted at the V_H-FR1. In silico protein modeling of scFv#m1-10, #m2-97, and the wt-scFv docked with cortisol was performed, but, the substitutions did not cause prominent alterations in the backbone and CDR conformations.⁷⁸⁾

5. SEEKING “HOT REGIONS” FOR MUTAGENESIS FOR MORE EFFICIENT ANTIBODY BREEDING

5.1. Affinity Maturations Based on Mutagenesis Targeting V_H-FR1

As shown in Section 4.4, many of the CAP-derived mutants with improved affinities contained substitutions and insertions in V_H-FR1. However, FRs have been considered to construct a β-sheet sandwich that supports six CDR loops, and thus do not directly affect the interaction with antigens. Therefore, this region was not mutagenized with the aim of affinity maturations. Thus, we performed two “designed” site-directed mutagenesis (A and B) on N-terminal partial positions of the V_H-FR1 in the aforementioned anti-cortisol wt-scFv⁸⁴⁾ (Fig. 11A). Mutagenesis A designed a limited set of amino acid substitutions at positions 1–3, 5–7, 9, and 10 to generate only 1536 variations, while mutagenesis B inserted 1–6 consecutive randomized amino acids between positions 6 and 7, to follow the unexpected but favorably functioned insertion observed in scFv#m1-2 shown above.

Fig. 11. Structures and Antigen-Binding Characteristics of Improved scFvs Generated via V_H-FR1-Targeted Mutagenesis [reproduced from Sci. Rep., 11, 8201 (2021), Springer Nature, with partial modifications]

(A) Design of mutagenesis A (left) and B (right). Mutagenesis A generates scFvs whose codons for amino acid positions 1–3, 5–7, 9, and 10 were degenerated to encode 2–6 predefined residues, whereas mutagenesis B generates scFvs in which extra 1–6 amino acid residue(s) were inserted between positions 6 and 7 using (NNS)_n degenerate codons^84,86,87) (n = 1–6). (B) Structures and K_a values⁹⁹⁾ of improved scFvs. Amino acid sequences at positions 1–10 of the V_H of scFvs obtained from mutagenesis A (left) and amino acid insertions of scFvs obtained from mutagenesis B (right). Newly incorporated amino acids are shown in red. Regarding the scFvs that showed the K_a exceeding 10¹⁰ M⁻¹, the K_a values were redetermined in triplicate and shown as mean ± S.D. (C) Typical dose–response curves for cortisol in competitive ELISAs (Fig. 1A-b) using wt-scFv and selected improved scFvs are shown with the midpoint values. Vertical bars indicate the S.D. (n = 4).

Screening the resulting scFv–phage libraries with CAP revealed various mutants with improved affinities (Fig. 11B). Seven scFvs were obtained via the mutagenesis A. The K_a values of these mutants ranged 0.64–1.1 × 10¹⁰ M⁻¹; thus 18–31-fold enhancement was achieved. All these mutants used Q and glutamic acid (E) for mutagenized positions 6 and 10, and many of them used E and S for positions 1 and 7, respectively. These residues may contribute to the formation of a motif with higher affinities. From the mutagenesis B, we isolated 14 improved scFvs showing the K_a ranging 0.62–1.1 × 10¹⁰ M⁻¹, corresponding to 17–31-fold enhancement. Species with a single amino acid insertion were the most numerous, but species with 2–6 residue insertions were all present. Mutant mB1#130 with a single insertion of aspartic acid (D), as well as mutant mB4#3 with four inserted residues, L, Q, L, and tryptophan (W), showed the highest affinity (K_a, 1.1 × 10¹⁰ M⁻¹).

The seven scFvs with the K_a values of 10¹⁰ M⁻¹ order, obtained from the mutagenesis A and B, exhibited significantly improved sensitivity in the competitive ELISA (Fig. 1A-b): scFvs mA#5 and mB1#129 generated dose–response curves with approx. 24-fold smaller midpoints (29.9 and 29.4 pg/assay, respectively) than those of wt-scFv (706 pg/assay) (Fig. 11C). The mutant mB1#130 that showed the largest K_a resulted in a slightly larger midpoint (37.5 pg/assay); this might be attributable to the “bridging phenomena.”⁸⁵⁾ The LOD of ELISAs using scFvs mA#5 and mB1#129 was 3.9 and 10.2 pg/assay, respectively, corresponding to approx. 2 ng/mL serum cortisol levels. Thus, these ELISAs are sensitive enough to measure the normal minimum levels of serum cortisol (see Section 4.3). A cross-reaction study allowed us to evaluate that these scFvs are specific for practical use.⁸⁴⁾

5.2. Improved CAP System for Isolating Mutants Showing Higher Affinities

As described in Section 5.1, we generated various scFvs with improved affinities; however, their K_a did not exceed 1.1 × 10¹⁰ M⁻¹. We assumed that there might be a limitation in recovering tightly bound scFv–phages from the microwells under acidic dissociation conditions (see Section 4.1).⁷⁸⁾ Therefore, we improved the CAP system by employing a dissociation-independent recovery method and used it to search the libraries obtained via the mutagenesis B (see Section 5.1).^86,87) According to the original CAP, monoclonal scFv–phages generated in each microwell reacted to the immobilized cortisol residues; however, in this system, the cortisol residues were immobilized via a linker containing a disulfide bond. Next, the scFv–phages with ordinarily high, moderate, and low affinities were eliminated using serial acidic and basic treatments. Subsequently, dissociation-resistant scFv–phages that remained bound to cortisol moieties were recovered via disulfide bond cleavage using dithiothreitol^88,89) [method b in Fig. 9(6)].

As expected, this system allowed us to isolate 24 scFvs that exhibited K_a values ranging 1.2–2.2 × 10¹⁰ M⁻¹, thereby exceeding the target value of 1.1 × 10¹⁰ M⁻¹. The scFv species with the top three K_a had the insertions as follows: #n1-260 (2.2 × 10¹⁰ M⁻¹), L; #n4-335 (2.0 × 10¹⁰ M⁻¹), AFAD; and #n6-52 (2.0 × 10¹⁰ M⁻¹) QVETTS [A, alanine and F, phenylalanine]. These scFvs generated significantly more sensitive dose–response curves with 44–68-fold smaller midpoints than the wt-scFv.⁸⁷⁾ Fusion proteins directly linking these scFvs with NanoLuc luciferase²⁴⁾ enabled bioluminescent ELISA systems (Fig. 1A-b) with further increased sensitivities, as shown by over 150-fold smaller midpoints. Because of their excellent sensitivity, serum specimens could be applied the ELISA after over 100-fold dilution. In the determination of control sera, acceptable values were obtained with satisfactory parallelism between dilution rates.⁸⁷⁾

5.3. Affinity Maturation Based on Systematic Exploration of V_H-CDR3

As described in Section 3.2, we generated an anti-E₂ scFv mutant (scFv#M3rd) showing 140-fold improved affinity over the wt-scFv after a three-step mutagenesis. This mutant had 11 substitutions, but scFv#4mut, which showed an even higher affinity (174-fold-improvement), was generated with only four out of the 11 substitutions⁷⁴⁾ (Fig. 12A). Furthermore, we determined that the substitution of L100gQ occurred in the V_H-CDR3 was the most important factor. In fact, restoration of the V_H-L100gQ to the original amino acid (i.e., substitution from Q to L) resulted in the most drastic reduction in the affinity (down to 2.3% of the scFv#4mut as the K_a), compared with the reduction due to restoration of one of the remaining three substitutions, I29V, L36M, and S77G [I, isoleucine and M, methionine] in the V_L (29, 34, and 63%, respectively).⁷⁴⁾ With another perspective, the V_H-L100gQ alone enhanced the affinity of the wt-scFv by 17-fold as shown with the scFv#R5-1, and the three V_L substitutions cooperatively assisted achieving the 174-fold increased affinity, as shown with the scFv#4mut (Fig. 12A).

Fig. 12. Summary of Antibody-Breeding for a scFv against E₂ Based on the Systematic Exploration of V_H-CDR3 [reproduced from Biol. Pharm. Bull. Vol. 45 No. 7 pages 851–855 (2022) with partial modifications]

(A) Outline of our previous antibody-breeding on anti-E₂ scFvs,^70,71,74) a part of which is described in Section 3.2. Stars represent amino acid substitutions, details of which in scFv#M3rd are shown in Fig. 8(A). Orange arrows indicate increases in E₂-binding affinity (K_a), and magnitudes are shown beside the arrows. (B) Concept of the two-stage affinity maturation. V_H-CDR3 amino acid sequence of wt-scFv is also shown. (C) Illustration of stepwise improvement in the scFv affinity. The introduced amino acid substitutions are denoted by stars and one-letter codes. Orange arrows indicate increases in K_a, and the magnitudes are shown beside the arrows. The K_a values (M⁻¹) of scFvs newly appeared in this figure are as follows: scFv#m1P, 5.9 × 10⁸; scFv#m2, 5.6 × 10⁹; scFv#m3, 3.0 × 10⁹; scFv#m4, 7.7 × 10⁹; and scFv#m7, 3.2 × 10¹⁰.

Considering these findings, we devised a systematic approach composed of two stages, each of which deals with small libraries⁹⁰⁾ (Fig. 12B). The first stage is the systematic exploration of substitutions in V_H-CDR3 that decisively increase affinity, and the second stage aims to further enhance the affinity of the first-stage products by adding extra substitutions in the entire scFv sequence. To evaluate the efficacy of this approach, we performed antibody-breeding of anti-E₂ scFv. We initially searched for other effective substitutions in V_H-CDR3 (Fig. 12B) by exploring 13 mini-libraries, each of which was randomized at one of the positions 95–100 g in the V_H (highly conserved residues 101 and 102 were excluded) to make any of the 20 proteinogenic amino acids appeared. However, we did not find any substitutions that were more effective than L100gQ (Fig. 12C-a). The second most effective substitution was that of histidine (H) to proline (P) at position 99. The resulting mutant scFv#m1P showed a 6.9-fold higher affinity over the wt-scFv. These L100gQ and H99P substitutions functioned cooperatively, as observed in scFv#m2 that displayed 65-fold higher affinity. Next, we explored a library generated by randomizing serial triple amino acids (e.g., 95–97 residues), which potentially contained 20³ scFv species (Fig. 12C-b). Comprehensive searching with the CAP system disclosed the serial substitutions E100eN–I100fA–L100gQ (N, asparagine) expressed in scFv#m3, which caused 35-fold enhanced affinity. Fortunately, the effect of this triplet was magnified by 2.6-fold by cooperation of the H99P, as shown with scFv#m4. Consequently, we found decisive substitution patterns in V_H-CDR3 i.e., H99P/E100eN/I100fA/L100gQ.

In the second step for “overwriting” fine-tunable substitutions, the error-prone PCR would be available. However, we tested the cooperativity of “the V_L substitution cluster” (I29V/L36M/S77G) that we found previously (Fig. 12A). This synergistically functioned and generated scFv#m7 that displayed the largest K_a (3.2 × 10¹⁰ M⁻¹), 372-fold larger than that of the wt-scFv. To the best of our knowledge, this improvement in affinity is the highest reported to date for mutagenesis targeting anti-steroid antibodies.

Because of the potential of V_H-CDR3 for antigen recognition, at the beginning of antibody engineering, thorough randomization was often performed targeting multiple (sometimes all) amino acids therein, generating vast libraries of mutants. These approaches were often effective in generating different specificities,^91,92) however, not always successful in improving the affinity while maintaining the original specificity. The current study encourages the re-evaluation of the utility of mutagenesis targeting V_H-CDR3 for affinity maturation. We believe that the current success was due to the potential of the CAP system for discovering improved mutants. The scFv#m7 enabled 25-fold more sensitive ELISA than that with the wt-scFv. Dose–response curves covered approx. 1.0–50 pg/assay with midpoint and the LOD of 4.5 and 0.78 pg/assay, respectively. Cross-reactivity with estrone, estriol, ethynylestradiol, progesterone, and cortisol was 0.53, 13, 3.8, 0.001, and <0.001%, respectively, which was satisfactorily low for clinical use.⁹⁰⁾

6. CONCLUSION: FUTURE PERSPECTIVES

It has been 45 years since the author started studying immunoassays for a bachelor’s degree. During that time, two great breakthroughs were achieved in antibody production, both of which led to the award of the Nobel prize.^13,60) First, the hybridoma technology enabled the production of mAbs,¹³⁾ which was followed by antibody engineering that generated “recombinant” antibodies with artificial structures.⁶⁰⁾ Fortunately, the author was blessed with opportunities to incorporate these leading-edge techniques into his studies directed to advanced immunoassays.

The hybridoma technology was established long ago as the standard method for producing “reagent-grade” native antibodies, which contributed to ensuring reproducibility of assay performances. Because myeloma cells available for cell fusion are mouse-derived, most mAbs have been prepared via the immortalization of antibody-producing cells generated in immunized mice by fusion with myeloma cells. However, it has been pointed out that, particularly against haptens, rabbits tend to generate antibodies with higher affinities.^93–95) Overcoming years of difficulty, rabbit-derived myeloma cell lines suitable for fusion with rabbit antibody-producing cells have recently been established.⁹⁴⁾ Hybridoma-based rabbit mAbs might be used widely hereafter.

Engineered antibodies, which are “artificial mAbs” with potentially higher antigen-binding performances, have been employed in commercial diagnostic products,⁹⁶⁾ but are still under developing status. In this review, we showed that our antibody-breeding approach significantly improved the affinity of hybridoma-derived wt-scFvs by 10-fold or more and consequently improved the sensitivity of immunoassays. We emphasize that the 100-fold improvement of sensitivity in competitive immunoassays described in Section 3.2, is almost impossible to achieve via conventional approaches, e.g., employing different signal-generating groups or detection methods thereof. By combining with the CAP system, the efficiency of antibody-breeding was drastically improved to generate extensive affinity-matured mutants with considerably minimal trial-and-error processes. Indeed, we obtained anti-cortisol and anti-E₂ scFvs showing 63- and 372-fold improved affinity (K_a, 2.4 × 10¹⁰ and 3.2 × 10¹⁰ M⁻¹; Sections 4.2 and 5.3), respectively. These affinities could be evaluated at the highest level that can be expected for native anti-hapten antibodies (discussed in Section 3.1), but have not reached to the affinity that we expected for artificial “future antibodies.” To the best of our knowledge, the most successful in vitro affinity maturation reported so far is the study performed on an scFv specific to a biotin-labeled fluorescein.⁹⁷⁾ Therein, random mutagenesis was repeated several times on a mouse-derived parental scFv, and the resulting mutant with 14 substitutions (12 of which were in the V_H) showed an extraordinarily high affinity (K_a = 3.7 × 10¹² M⁻¹) via 2600-fold improvement. Our studies on the anti-E₂ scFv, referred to above, would be evaluated as the second-best successful performance in affinity maturation targeting anti-hapten antibodies. Hereafter, it will be necessary to generate mutants with an affinity exceeding 10¹¹ (M⁻¹) order, which should enable subpicomolar measurements of analytes (e.g., less than 0.1 pg/mL of steroids). To avoid overlooking rare mutants in a vast library, the throughput of the CAP system should be enhanced to analyze an entire library composed of over 10⁶ members.

Throughout our study, we followed the strategy to improve scFvs derived from hybridoma mAbs (wt-scFvs) that have already been directed to a desired specificity by immunization, though it was also worth challenging to improve scFvs selected from “naïve” or “synthetic” libraries.^59,61–63) As far as we experienced, it is unlikely that the specificities of scFvs change drastically due to the introduction of a limited number of amino acid substitutions that are effective for affinity maturation. Therefore, parental mAbs should be as specific as possible to the target antigens. If the target analyte is a hapten, immunization with carefully designed hapten–carrier conjugates, as discussed in Section 2.1, should provide suitable starting materials.

Rabbit mAbs are attractive tools for establishing hapten immunoassays with excellent sensitivity. Genetic engineering has served as a practical method to produce rabbit antibody fragments via the cloning of rabbit V_H and V_L genes using peripheral blood cells.^93–95) We are interested in antibody-breeding on rabbit scFvs as a new challenge for developing more sensitive hapten immunoassays. Compared with mouse antibodies, rabbit antibodies are characterized by longer V_L-CDR3 and greater contribution of the V_L domains to the interaction with antigens.⁹⁵⁾ V_L-CDR3-focused mutagenesis might be a promising strategy for generating affinity-matured mutants.

As is well recognized, scFvs can be genetically modified into a variety of derivatives, e.g., fusion proteins with various functional molecules. We showed that scFv-luciferase fusions facilitate highly sensitive assays in both competitive and noncompetitive formats,²⁴⁾ and developed a bioluminescent immunoassay for cortisol, providing dose–response curves with over 150-fold decreased midpoints (see Section 5.2).⁸⁷⁾ We expect that these engineered antibodies will soon become popular, opening the door to next-generation immunoassays that enable simpler and more reliable determination of bioactive compounds.

Acknowledgments

I would like to express my deepest gratitude to Dr. Toshio Nambara, Emeritus Professor of Tohoku University, for his warm-hearted guidance and encouragement. I sincerely thank Dr. Hiroshi Hosoda, former Associate Professor of Tohoku University; Dr. Kazutake Shimada, Emeritus Professor of Kanazawa University; and Dr. Junichi Goto, Emeritus Professor of Tohoku University, for their kind guidance and support. I would also like to thank Drs. Tadao Terao and Jun-ichi Sawada, former Director General and Director of the National Institute of Health Sciences, for their kind guidance on hybridoma technology. My thanks are also due to Drs. Carl Borrebaeck and Eskil Söderlind, Professor and Associate Professor of Lund University, for their kind guidance on antibody engineering. I appreciate the cooperation of Drs. Hiroyuki Oyama, Izumi Morita, and Yuki Kiguchi of Kobe Pharmaceutical University. Finally, I would like to express my gratitude to Dr. Kazuhiro Imai, Visiting Professor of Musashino University, Professor Emeritus of the University of Tokyo who kindly nominated me for “The Pharmaceutical Society of Japan Award for Divisional Scientific Contribution,” and Drs. Akio Tsuji and Hidetoshi Arakawa, Emeritus Professors of Showa University, for their warm encouragement.

Conflict of Interest

The author declares no conflict of interest.

Notes

This review of the author’s work was written by the author upon receiving the 2024 Pharmaceutical Society of Japan Award for Divisional Scientific Contribution.

REFERENCES

• 1) Berson SA, Yalow RS. Quantitative aspects of the reaction between insulin and insulin-binding antibody. J. Clin. Invest., 38, 1996–2016 (1959).
• 2) Yalow RS, Berson SA. Assay of plasma insulin in human subjects by immunological methods. Nature, 184 (Suppl. 21), 1648–1649 (1959).
• 3) Glick S. Rosalyn Sussman Yalow (1921–2011). The second woman to win the Nobel prize in medicine. Nature, 474, 580 (2011).
• 4) Wild D ed. The Immunoassay Handbook: Theory and applications of ligand binding, ELISA and related techniques, 4th Edition. Elsevier, the Netherlands (2013).
• 5) The Immunoassay: basics and applications. (Kobayashi N, Ueda H, Miyake S, Arakawa H eds.) Kodansha, Tokyo (2014).
• 6) Engvall E, Perlmann P. Enzyme-linked immunosorbent assay (ELISA). Quantitative assay of immunoglobulin G. Immunochemistry, 8, 871–874 (1971).
• 7) Lateral flow immunoassay. (Wong R, Tse H eds.) Humana Press, New York, NY, U.S.A. (2009).
• 8) Kobayashi N, Goto J, Shimada K, Matsuki Y, Kato Y. Generation of high performance anti-hapten monoclonal antibodies and their application to trace characterization. Jpn. J. Clin. Chem., 34, 125–145 (2005).
• 9) Samarajeewa U, Wei CI, Huang TS, Marshall MR. Application of immunoassay in the food industry. Crit. Rev. Food Sci. Nutr., 29, 403–434 (1991).
• 10) Sherry J. Environmental immunoassays and other bioanalytical methods: overview and update. Chemosphere, 34, 1011–1025 (1997).
• 11) Kobayashi N, Goto J. Noncompetitive immunoassays for small molecules with high sensitivity and specificity. Adv. Clin. Chem., 36, 139–170 (2001).
• 12) Köhler G, Milstein C. Continuous cultures of fused cells secreting antibody of predefined specificity. Nature, 256, 495–497 (1975).
• 13) Leavy O. The birth of monoclonal antibodies. Nat. Immunol., 17 (S1), S13 (2016).
• 14) Kobayashi N, Sun P, Fujimaki Y, Niwa T, Nishio T, Goto J, Hosoda H. Generation of a novel monoclonal antibody against cortisol-[C-4]-bovine serum albumin conjugate: application to enzyme-linked immunosorbent assay for urinary and serum cortisol. Anal. Sci., 18, 1309–1314 (2002).
• 15) Kobayashi N, Kato Y, Oyama H, Taga S, Niwa T, Sun P, Ohtoyo M, Goto J. Anti-estradiol-17β single-chain Fv fragments: generation, characterization, gene randomization, and optimized phage display. Steroids, 73, 1485–1499 (2008).
• 16) Oyama H, Morita I, Kiguchi Y, Miyake S, Moriuchi A, Akisada T, Niwa T, Kobayashi N. Gaussia luciferase as a genetic fusion partner with antibody fragments for sensitive immunoassay monitoring of clinical biomarkers. Anal. Chem., 87, 12387–12395 (2015).
• 17) Kobayashi N, Oiwa H, Goto J. Production and characterization of group-specific monoclonal antibodies recognizing nonamidated, glycine- and taurine-amidated ursodeoxycholic acid 7-N-acetylglucosaminides. J. Steroid Biochem. Mol. Biol., 64, 171–177 (1998).
• 18) Kobayashi N, Katsumata H, Katayama H, Oiwa H, Goto J, Takeuchi Y. A monoclonal antibody-based enzyme-linked immunosorbent assay of ursodeoxycholic acid 3-sulfates in human urine. J. Steroid Biochem. Mol. Biol., 72, 265–272 (2000).
• 19) Kobayashi N, Katsumata H, Uto Y, Goto J, Niwa T, Kobayashi K, Mizuuchi Y. A monoclonal antibody-based enzyme-linked immunosorbent assay of glycolithocholic acid sulfate in human urine for liver function test. Steroids, 67, 827–833 (2002).
• 20) Kobayashi N, Hisada A, Shimada K. Specificity of the polyclonal antibodies raised against a novel 25-hydroxyvitamin D₃-bovine serum albumin conjugate linked through the C-11α position. J. Steroid Biochem. Mol. Biol., 48, 567–572 (1994).
• 21) Kobayashi N, Imazu T, Kitahori J, Mano H, Shimada K. A selective immunoaffinity chromatography for determination of plasma 1α,25-dihydroxyvitamin D₃: application of specific antibodies raised against a 1α,25-dihydroxyvitamin D₃–bovine serum albumin conjugate linked through the 11α-position. Anal. Biochem., 244, 374–383 (1997).
• 22) Kobayashi N, Imazu T, Ebisawa A, Shimada K. Production and characterization of monoclonal antibodies against a novel 1α,25-dihydroxyvitamin D₃-bovine serum albumin conjugate linked through the 11α-position. J. Steroid Biochem. Mol. Biol., 63, 127–137 (1997).
• 23) Higashi T, Kobayashi N, Ohmi H, Hayashi Y, Shimada K. Enzyme-linked immunosorbent assay for plasma 24,25-dihydroxyvitamin D₃. Anal. Chim. Acta, 365, 151–158 (1998).
• 24) Oyama H, Kiguchi Y, Morita I, Miyashita T, Ichimura A, Miyaoka H, Izumi A, Terasawa S, Osumi N, Tanaka H, Niwa T, Kobayashi N. NanoLuc luciferase as a suitable fusion partner of recombinant antibody fragments for developing sensitive luminescent immunoassays. Anal. Chim. Acta, 1161, 238180 (2021).
• 25) Okuyama M, Kobayashi N, Takeda W, Anjo T, Matsuki Y, Goto J, Kambegawa A, Hori S. Enzyme-linked immunosorbent assay for monitoring toxic dioxin congeners in milk based on a newly generated monoclonal anti-dioxin antibody. Anal. Chem., 76, 1948–1956 (2004).
• 26) Kobayashi N, Banzono E, Shimoda Y, Oyama H, Kunihiro T, Morita I, Ohta M. A monoclonal antibody-based enzyme-linked immunosorbent assay for human urinary cotinine to monitor tobacco smoke exposure. Anal. Methods, 3, 1995–2002 (2011).
• 27) Morita I, Oyama H, Yasuo M, Matsuda K, Katagi K, Ito A, Tatsuda H, Tanaka H, Morimoto S, Kobayashi N. Antibody fragments for on-site testing of cannabinoids generated via in vitro affinity maturation. Biol. Pharm. Bull., 40, 174–181 (2017).
• 28) Morita I, Oyama H, Kanda Y, Yasuo M, Ito A, Toyota M, Hayashi Y, Yokoyama T, Kobayashi N. Enantioselective monoclonal antibodies for detecting ketamine to crack down on illicit use. Biol. Pharm. Bull., 41, 123–131 (2018).
• 29) Morita I, Oyama H, Kiguchi Y, Oguri A, Fujimoto N, Takeuchi A, Tanaka R, Ogata J, Kikura-Hanajiri R, Kobayashi N. Immunochemical monitoring of psilocybin and psilocin to identify hallucinogenic mushrooms. J. Pharm. Biomed. Anal., 190, 113485 (2020).
• 30) Morita I, Kiguchi Y, Oyama H, Yamaki K, Sakio N, Kashiwabara K, Kuroda Y, Ito A, Yokota A, Ikeda N, Kikura-Hanajiri R, Ueda H, Numazawa S, Yoshida T, Kobayashi N. Derivatization-assisted immunoassays: application for group-specific detection of potent methamphetamine and amphetamine enantiomers. Anal. Methods, 14, 2745–2753 (2022).
• 31) Mawer EB. Clinical implications of measurements of circulating vitamin D metabolites. Clin. Endocrinol. Metab., 9, 63–79 (1980).
• 32) Shimada K, Kobayashi N. Recent advances in the analysis of vitamin D metabolites. TrAC Trends Anal. Chem., 10, 103–105 (1991).
• 33) Porteous CE, Coldwell RD, Trafford DJH, Makin HLJ. Recent developments in the measurement of vitamin D and its metabolites in human body fluids. J. Steroid Biochem., 28, 785–801 (1987).
• 34) Kobayashi N, Kitahori J, Mano H, Shimada K. Syntheses of 11α-(3-carboxypropanoyloxy)-1α,25-dihydroxyvitamin D₃ and 11α-(4-carboxybutanoyloxy)-1α,25-dihydroxyvitamin D₃: novel haptenic derivatives for production of highly specific antibodies to 1α,25-dihydroxyvitamin D₃. J. Chem. Soc., Perkin Trans. 1, 1994, 1809–1815 (1994).
• 35) Marschall H-U, Egestad B, Matern H, Matern S, Sjövall J. N-Acetylglucosaminides: a new type of bile acid conjugate in man. J. Biol. Chem., 264, 12989–12993 (1989).
• 36) Niwa T, Fujita K, Goto J, Nambara T. Separation and characterization of ursodeoxycholate 7-N-acetylglucosaminides in human urine by high-performance liquid chromatography with fluorescence detection. J. Liq. Chromatogr., 16, 2531–2544 (1993).
• 37) Nakamura K, Yoneda M, Kimura A, Tamori K, Yokohama S, Sato Y, Kato T, Hasegawa T, Saito H, Aoshima M, Fujita M, Makino I. Increase of sulfated ursodeoxycholic acid in the serum and urine of patients with chronic liver disease after ursodeoxycholic acid therapy. J. Gastroenterol. Hepatol., 11, 385–390 (1996).
• 38) Tylš F, Páleníček T, Horáček J. Psilocybin – summary of knowledge and new perspectives. Eur. Neuropsychopharmacol., 24, 342–356 (2014).
• 39) Dinis-Oliveira RJ. Metabolism of psilocybin and psilosin: clinical and forensic toxicological relevance. Drug Metab. Rev., 49, 84–91 (2017).
• 40) Waters R. Methamphatamine & other amphetamines, Mason Crest, Folcroft, PA, USA, 2014.
• 41) Wang T, Shen B, Wu H, Hu J, Xu H, Shen M, Xiang P. Disappearance of R/S-methamphetamine and R/S-amphetamine from human scalp hair after discontinuation of methamphetamine abuse. Forensic Sci. Int., 284, 153–160 (2018).
• 42) Li L, Everhart T, Jacob P III, Jones R, Mendelson J. Stereoselectivity in the human metabolism of methamphetamine. Br. J. Clin. Pharmacol., 69, 187–192 (2010).
• 43) Shute RE, Rich DH. Synthesis and evaluation of novel activated mixed carbonate reagents for the introduction of the 2-(trimethylsilyl)ethoxycarbonyl(Teoc)-protecting group. Synthesis (Stuttg.), 1987, 346–349 (1987).
• 44) Kobayashi N, Oyama H. Antibody engineering toward high-sensitivity high-throughput immunosensing of small molecules. Analyst, 136, 642–651 (2011).
• 45) Chappey ON, Sandouk P, Scherrmann JMG. Monoclonal antibodies in hapten immunoassays. Pharm. Res., 9, 1375–1379 (1992).
• 46) Chappey O, Debray M, Niel E, Scherrmann JM. Association constants of monoclonal antibodies for hapten: heterogeneity of frequency distribution and possible relationship with hapten molecular weight. J. Immunol. Methods, 172, 219–225 (1994).
• 47) Ullman EF, Milburn G, Jelesko J, Radika K, Pirio M, Kempe T, Skold C. Anti-immune complex antibodies enhance affinity and specificity of primary antibodies. Proc. Natl. Acad. Sci. U.S.A., 90, 1184–1189 (1993).
• 48) Self CH, Dessi JL, Winger LA. High-performance assays of small molecules: enhanced sensitivity, rapidity, and convenience demonstrated with a noncompetitive immunometric anti-immune complex assay system for digoxin. Clin. Chem., 40, 2035–2041 (1994).
• 49) Towbin H, Motz J, Oroszlan P, Zingel O. Sandwich immunoassay for the hapten angiotensin II. A novel assay principle based on antibodies against immune complexes. J. Immunol. Methods, 181, 167–176 (1995).
• 50) Pulli T, Höyhtyä M, Söderlund H, Takkinen K. One-step homogeneous immunoassay for small analytes. Anal. Chem., 77, 2637–2642 (2005).
• 51) Leivo J, Kivimäki L, Juntunen E, Pettersson K, Lamminmäki U. Development of anti-immunocomplex specific antibodies and non-competitive time-resolved fluorescence immunoassay for the detection of estradiol. Anal. Bioanal. Chem., 411, 5633–5639 (2019).
• 52) Ozeki Y, Tanimura Y, Nagai S, Nomura T, Kinoshita M, Shibuta K, Matsuda N, Miyamoto S, Yoshida Y, Okamoto M, Gotoh K, Masaki T, Kambara K, Shibata H. Development of a new chemiluminescent enzyme immunoassay using a two-step sandwich method for measuring aldosterone concentrations. Diagnostics (Basel), 11, 433 (2021).
• 53) Barnard G, Kohen F. Idiometric assay: noncompetitive immunoassay for small molecules typified by the measurement of estradiol in serum. Clin. Chem., 36, 1945–1950 (1990).
• 54) Kobayashi N, Oiwa H, Kubota K, Sakoda S, Goto J. Monoclonal antibodies generated against an affinity-labeled immune complex of an anti-bile acid metabolite antibody: an approach to noncompetitive hapten immunoassays based on anti-idiotype or anti-metatype antibodies. J. Immunol. Methods, 245, 95–108 (2000).
• 55) Kobayashi N, Shibusawa K, Kubota K, Hasegawa N, Sun P, Niwa T, Goto J. Monoclonal anti-idiotype antibodies recognizing the variable region of a high-affinity antibody against 11-deoxycortisol. Production, characterization and application to a sensitive noncompetitive immunoassay. J. Immunol. Methods, 274, 63–75 (2003).
• 56) Niwa T, Kobayashi T, Sun P, Goto J, Oyama H, Kobayashi N. An enzyme-linked immunometric assay for cortisol based on idiotype-anti-idiotype reactions. Anal. Chim. Acta, 638, 94–100 (2009).
• 57) Kobayashi N, Iwakami K, Kotoshiba S, Niwa T, Kato Y, Mano N, Goto J. Immunoenzymometric assay for a small molecule, 11-deoxycortisol, with attomole-range sensitivity employing an scFv-enzyme fusion protein and anti-idiotype antibodies. Anal. Chem., 78, 2244–2253 (2006).
• 58) Winter G, Milstein C. Man-made antibodies. Nature, 349, 293–299 (1991).
• 59) Bradbury ARM, Sidhu S, Dübel S, McCafferty J. Beyond natural antibodies: the power of in vitro display technologies. Nat. Biotechnol., 29, 245–254 (2011).
• 60) Gibney E, Noorden RV, Ledford H, Castelvecchi D, Warren M. ‘Test-tube’ evolution wins Chemistry Nobel Prize. Nature, 562, 176 (2018).
• 61) Chiu ML, Goulet DR, Teplyakov A, Gilliland GL. Antibody structure and function: the basis for engineering therapeutics. Antibodies (Basel), 8, 55 (2019).
• 62) Introduction to Antibody Engineering. (Rüker F, Wozniak-Knopp G eds.) Springer, Germany (2021).
• 63) Stone CA Jr, Spiller BW, Smith SA. Engineering therapeutic monoclonal antibodies. J. Allergy Clin. Immunol., 153, 539–548 (2024).
• 64) Smith GP, Petrenko VA. Phage display. Chem. Rev., 97, 391–410 (1997).
• 65) Skerra A, Plückthun A. Assembly of a functional immunoglobulin Fv fragment in Escherichia coli. Science, 240, 1038–1041 (1988).
• 66) Bird RE, Hardman KD, Jacobson JW, Johnson S, Kaufman BM, Lee SM, Lee T, Pope SH, Riordan GS, Whitlow M. Single-chain antigen-binding proteins. Science, 242, 423–426 (1988).
• 67) Phage Display. (Clackson T, Lowman HB eds.) Oxford University Press, Oxford, U.K. (2004).
• 68) Frenzel A, Kügler J, Helmsing S, Meier D, Schirrmann T, Hust M, Dübel S. Designing human antibodies by phage display. Transfus. Med. Hemother., 44, 312–318 (2017).
• 69) Leung DW, Chen E, Goeddel DV. A method for random mutagenesis of a defined DNA segment using a modified polymerase chain reaction. Technique, 1, 11–15 (1989).
• 70) Kobayashi N, Oyama H, Kato Y, Goto J, Söderlind E, Borrebaeck CAK. Two-step in vitro antibody affinity maturation enables estradiol-17β assays with more than 10-fold higher sensitivity. Anal. Chem., 82, 1027–1038 (2010).
• 71) Oyama H, Yamaguchi S, Nakata S, Niwa T, Kobayashi N. “Breeding” diagnostic antibodies for higher assay performance: a 250-fold affinity-matured antibody mutant targeting a small biomarker. Anal. Chem., 85, 4930–4937 (2013).
• 72) Oyama H, Morita I, Kiguchi Y, Banzono E, Ishii K, Kubo S, Watanabe Y, Hirai A, Kaede C, Ohta M, Kobayashi N. One-shot in vitro evolution generated an antibody fragment for testing urinary cotinine with more than 40-fold enhanced affinity. Anal. Chem., 89, 988–995 (2017).
• 73) Oyama H, Morita I, Kiguchi Y, Morishita T, Fukushima S, Nishimori Y, Niwa T, Kobayashi N. A single-step “breeding” generated a diagnostic anti-cortisol antibody fragment with over 30-fold enhanced affinity. Biol. Pharm. Bull., 40, 2191–2198 (2017).
• 74) Oyama H, Kiguchi Y, Morita I, Yamamoto C, Higashi Y, Taguchi M, Tagawa T, Enami Y, Takamine Y, Hasegawa H, Takeuchi A, Kobayashi N. Seeking high-priority mutations enabling successful antibody-breeding: systematic analysis of a mutant that gained over 100-fold enhanced affinity. Sci. Rep., 10, 4807 (2020).
• 75) Kabat EA, Wu TT, Perry HM, Gottesman KS, Foeller C. Sequences of Proteins of Immunological Interest; U. S. Department of Health and Human Services, National Institutes of Health, U. S. Government Printing Office, Washington, DC (1991).
• 76) Oliva B, Bates PA, Querol E, Avilés FX, Sternberg MJE. Automated classification of antibody complementarity determining region 3 of the heavy chain (H3) loops into canonical forms and its application to protein structure prediction. J. Mol. Biol., 279, 1193–1210 (1998).
• 77) D’Angelo S, Ferrara F, Naranjo L, Erasmus MF, Hraber P, Bradbury ARM. Many routes to an antibody heavy-chain CDR3: necessary, yet insufficient, for specific binding. Front. Immunol., 9, 395 (2018).
• 78) Kiguchi Y, Oyama H, Morita I, Morikawa M, Nakano A, Fujihara W, Inoue Y, Sasaki M, Saijo Y, Kanemoto Y, Murayama K, Baba Y, Takeuchi A, Kobayashi N. Clonal array profiling of scFv-displaying phages for high-throughput discovery of affinity-matured antibody mutants. Sci. Rep., 10, 14103 (2020).
• 79) Kiguchi Y, Oyama H, Morita I, Katayama E, Fujita M, Narasaki M, Yokoyama A, Kobayashi N. Antibodies and engineered antibody fragments against M13 filamentous phage to facilitate phage-display-based molecular breeding. Biol. Pharm. Bull., 41, 1062–1070 (2018).
• 80) Chrousos GP, Kino T, Charmandari E. Evaluation of the hypothalamic-pituitary-adrenal axis function in childhood and adolescence. Neuroimmunomodulation, 16, 272–283 (2009).
• 81) Crichton D, Grattage L, McDonald A, Corrie JET, Steel CM, Hubbard AL, Al-Dujaili EAS, Edwards CRW. The production and assessment of monoclonal antibodies to cortisol. Steroids, 45, 503–517 (1985).
• 82) Lewis JG, Manley L, Whitlow JC, Elder PA. Production of a monoclonal antibody to cortisol: application to a direct enzyme-linked immunosorbent assay of plasma. Steroids, 57, 82–85 (1992).
• 83) Oyama H, Tanaka E, Kawanaka T, Morita I, Niwa T, Kobayashi N. Anti-idiotype scFv-enzyme fusion proteins: a clonable analyte-mimicking probe for standardized immunoassays targeting small biomarkers. Anal. Chem., 85, 11553–11559 (2013).
• 84) Kiguchi Y, Oyama H, Morita I, Nagata Y, Umezawa N, Kobayashi N. The V_H framework region 1 as a target of efficient mutagenesis for generating a variety of affinity-matured scFv mutants. Sci. Rep., 11, 8201 (2021).
• 85) Hosoda H, Kobayashi N, Ishii N, Nambara T. Bridging phenomena in steroid immunoassays. The effect of bridge length on sensitivity in enzyme immunoassay. Chem. Pharm. Bull., 34, 2105–2111 (1986).
• 86) Kiguchi Y, Morita I, Tsuruno A, Kobayashi N. Retrieving dissociation-resistant antibody mutants: an efficient strategy for developing immunoassays with improved sensitivities. Biol. Pharm. Bull., 45, 1432–1437 (2022).
• 87) Kiguchi Y, Morita I, Yamaki K, Takegami S, Kobayashi N. Framework-directed amino-acid insertions generated over 55-fold affinity-matured antibody fragments that enabled sensitive luminescent immunoassays of cortisol. Biol. Pharm. Bull., 46, 1661–1665 (2023).
• 88) Kobayashi N, Karibe T, Goto J. Dissociation-independent selection of high-affinity anti-hapten phage antibodies using cleavable biotin-conjugated haptens. Anal. Biochem., 347, 287–296 (2005).
• 89) Kobayashi N, Oyama H, Nakano M, Kanda T, Banzono E, Kato Y, Karibe T, Nishio T, Goto J. “Cleavable” hapten-biotin conjugates: preparation and use for the generation of anti-steroid single-domain antibody fragments. Anal. Biochem., 387, 257–266 (2009).
• 90) Morita I, Kiguchi Y, Nakamura S, Yoshida A, Kubo H, Ishida M, Oyama H, Kobayashi N. More than 370-fold increase in antibody affinity to estradiol-17β by exploring substitutions in the V_H-CDR3. Biol. Pharm. Bull., 45, 851–855 (2022).
• 91) Barbas CF III, Bain JD, Hoekstra DM, Lerner RA. Semisynthetic combinatorial antibody libraries: a chemical solution to the diversity problem. Proc. Natl. Acad. Sci. U.S.A., 89, 4457–4461 (1992).
• 92) Griffiths AD, Williams SC, Hartley O, Tomlinson IM, Waterhouse P, Crosby WL, Kontermann RE, Jones PT, Low NM, Allison TJ, Prospero TD, Hoogenboom HR, Nissim A, Cox JPL, Harrison JL, Zaccolo M, Gherardi E, Winter G. Isolation of high affinity human antibodies directly from large synthetic repertoires. EMBO J., 13, 3245–3260 (1994).
• 93) Zhang Z, Liu H, Guan Q, Wang L, Yuan H. Advances in the isolation of specific monoclonal rabbit antibodies. Front. Immunol., 8, 494 (2017).
• 94) Chen Z, Wang G. Progress and perspectives of rabbit monoclonal antibodies. Blood Genom., 7, 13–21 (2023).
• 95) Li Y, Kong Y, Yu X, Yu W, Wen K, Shen J, Wang Z. Characteristics of rabbit hapten-specific and germline-based BCR repertoires following repeated immunization. One Health Adv., 1, 17 (2023).
• 96) Kobayashi N, Morita I. Immunoassays for biomarkers in clinical testing: advances and perspectives. Jpn. J. Clin. Chem., 53, 5–16 (2024).
• 97) Boder ET, Midelfort KS, Wittrup KD. Directed evolution of antibody fragments with monovalent femtomolar antigen-binding affinity. Proc. Natl. Acad. Sci. U.S.A., 97, 10701–10705 (2000).
• 98) Abraham GE. Solid-phase radioimmunoassay of estradiol-17β. J. Clin. Endocrinol. Metab., 29, 866–870 (1969).
• 99) Scatchard G. The attractions of proteins for small molecules and ions. Ann. N. Y. Acad. Sci., 51, 660–672 (1949).
• 100) Hosoda H, Sakai Y, Yoshida H, Miyairi S, Ishii K, Nambara T. The preparation of steroid N-hydroxysuccinimide esters and their reactivities with bovine serum albumin. Chem. Pharm. Bull., 27, 742–746 (1979).